4

I'm a beginner (started learning yesterday only) in modular arithmetic.

The question is to find the remainder when ${}^{72}C_{36}$ is divided 73 (where ${}^nC_r $ denotes ${n \choose k}$)

I know such problems can be answered by Lucas' theorem, but in this case, it's pointless.

I can't treat $(36!)^2$ as modular inverse (and using Wilson's identity) either because the number is huge. Same for Chinese remainder theorem.

And with the above 3 approaches, I'm out of options. No clue how to solve it, even the hint given isn't "good" (and I can't even prove the "hint")

Hint: ${72 \choose 36}={73\choose 0} + {73 \choose 1} +\cdots + {73\choose 36}$

Everything about this question, including the hint, is just bizzare to me! P. S. I don't want to use the hint (It's actually the complete solution)

And the hint is wrong.

Fawkes4494d3
  • 2,984
UmbQbify
  • 889
  • 2
    Can you show that if $1\leq k \leq p-1$ and $p$ is prime that $p | \binom{p}{k}$? – Integrand Jun 20 '20 at 20:32
  • By Luca's theorem? Yeah, it'll be obvious (since the unit digit in base p will always be greater unit digit of p in base p for any number less than p. I don't know of any other method (yet) – UmbQbify Jun 20 '20 at 20:38
  • 2
    What does it mean if a number is prime? What are its divisors? – Integrand Jun 20 '20 at 20:38
  • 1
    Is ${}^{72}C_{36} = \binom{72}{36}$? (If so, the hint is nonsense) – Brian Moehring Jun 20 '20 at 20:40
  • @Brian Moehring, yes, obviously. – UmbQbify Jun 20 '20 at 20:41
  • 1
    No, not by Luca's Theorem. By noting that ${p \choose k} = \frac {p!}{k!(p-k)!}$ and noting that $p|p!$ but $p\not \mid k!(p-k)!$. – fleablood Jun 20 '20 at 20:43
  • @Integrand, yeah, I get what you're trying to say, prime will always be there, that too just once. How does that help here( in ${72 \choose 36}$)? – UmbQbify Jun 20 '20 at 20:45
  • @fleablood, yeah, I was typing that. But how's it helpful? I mean $ {72\choose36}=\frac{37}{73}{73\choose36}$ but that doesn't seem to help – UmbQbify Jun 20 '20 at 20:48
  • 2
    "yes, obviously" That's obnoxious. Not all text use the same notation and it's not clear what the notation $^{72}C_{36}$ is supposed to mean. And as the hint is not true its very unclear what $^{72}C_{36}$ – fleablood Jun 20 '20 at 20:49
  • Well if the hint is true then $73|{73\choose k}$ so the remainder of that sum in the hint is .... – fleablood Jun 20 '20 at 20:51
  • @fleablood, well it was asked in test. So – UmbQbify Jun 20 '20 at 20:52
  • "well it was asked in test. So..." And are we all in the same class as you? Did we all take the same test? – fleablood Jun 20 '20 at 20:54
  • I meant, I can't do anything about the hint or the question. Besides, in my place the C notation is more popular – UmbQbify Jun 20 '20 at 20:59
  • 2
    As unsolicited advice, purge the word "obviously" from your mathematical vocabulary. It only served to ignore my point about the hint. I had seen both notations previously, but it is wholly unusual to see them both in the same text, even moreso in the same line. – Brian Moehring Jun 20 '20 at 21:08
  • @Brian Moehring, I didn't see the edit. I'm more used to C notation but it's longer to type it that way. I'll edit it. – UmbQbify Jun 20 '20 at 21:16
  • That and the hint is utter nonsense..... Although If the hint where true the solution would be easy. My and integrands solutions were not that obvious. – fleablood Jun 20 '20 at 21:27

3 Answers3

10

Just by the way, I think the hint was intended to alternate plus and minus:

${72 \choose 36}$ = ${73 \choose 0} - {73 \choose 1} + {73 \choose 2} - ... + {73 \choose 36}$

${\qquad\equiv 1} - 0 + 0 - ... + {0 \mod73}$ ${\qquad\equiv 1\mod73}$

as ${73}$ divides into ${73 \choose 1}$, ${73 \choose 2}$, ... , ${73 \choose 36}$, but not into ${73\choose 0}$.

... built on Jose Carlos Santos's answer, which he deleted.

wotnotv
  • 293
4

the hint is nonsense but

I noticed the denominator of $(36!)(36!)$ made me think that then numbers $1$ to $36$ are equiv $-72$ througe $-37\pmod {73}$ so $(36!)(36!)\equiv (36!)(-37)*(-38)*...*(-72) \equiv 72!(-1)^{36}\pmod {37}$ which made me realize the following result:

for any prime $p$, because $\mathbb Z_p$ is a field and every non-zero equivalence as an inverse:

$ {p-1\choose \frac {p-1}2}=\frac {(p-1)!}{(\frac {p-1}2!)^2}\equiv $

$(p-1)!\frac 1{1*2*.....*\frac {p-1}2}\frac 1{\frac {p-1}2*....*2*1}\equiv $

$(p-1)!\frac 1{1*2*.....*\frac {p-1}2}\frac 1{(-\frac {p+1}2)*....*(-2)*(-1)*(-1)^{\frac {p-1}2}}\equiv $

$(p-1)!\frac 1{1*2*......*\frac {p-1}2*\frac {p+1}2*....*(p-2)(p-1)(-1)^{\frac {p-1}2}}\equiv $

$(p-1)!\frac 1{(p-1)!(-1)^{\frac {p-1}2}}\equiv(-1)^{\frac {p-1}2}\pmod p$.

So $ {72 \choose 36} \equiv (-1)^{36}\equiv 1 \pmod {73}$

fleablood
  • 124,253
  • 2
    Yeah, this is the intelligent way to use Wilson's Theorem. – Integrand Jun 20 '20 at 21:18
  • Dont even need Wilson's theory. Just that refering to inverses and division makes sense in $\mathbb Z_p$. – fleablood Jun 20 '20 at 21:24
  • Wow, took me a while to understand. Amazing solution. Thanks! – UmbQbify Jun 20 '20 at 21:24
  • So actually, for any prime p, binomial coefficient of upper index p-1, the modulo is just $(-1)^{((p-1)-k)}$ for any valid lower index – UmbQbify Jun 20 '20 at 22:22
  • I was generalizing this, and it came out to be $$ { p-r \choose k} = (-1)^{(p-k-2r+1)}\dfrac{\displaystyle \prod_{i=1}^{r-1} (k+i)}{(r-1)!}$$ for $r\ge2$ – UmbQbify Jun 20 '20 at 22:26
  • I want to confirm something though, it's possible to substitute the modulo of the number only because they're coprime, right? – UmbQbify Jun 20 '20 at 22:29
  • I realised that there are more restrictions than they appear. $k<\frac{p-r}{2}$ but that's not a problem – UmbQbify Jun 20 '20 at 22:39
  • "So actually, for any prime p, binomial coefficient of upper index p-1, the modulo is just" Don't know about that. If $k=(p-1-k)$ this works becase $k!(p-1-k)!$ is "kind of" the same thing as $p!$. Wheres if $k < \frac {p-1}2$ then $k!(p-1-k)!$ will "kind of" be $p!$ with a hunk taken out the middle... Oh! but then $\frac{p!}{k!(p-1-k)!}\equiv (-1)^{2(\frac {p-1}2-k)}\prod_{j=\frac {p-1}2-k}^{\frac {p-1}2+k} j\pmod p$. ... I think. – fleablood Jun 20 '20 at 22:51
  • 1
    Oh... I guess that's what you got too..... – fleablood Jun 20 '20 at 22:52
  • I didn't understand anything at all. @fleablood. – UmbQbify Jun 20 '20 at 23:27
3

Not a very insightful answer, but sometimes there's nothing wrong with getting your hands dirty.

Since $73$ is prime, by Wilson's Theorem $72!\equiv -1 \bmod 73$. Let's try and compute $(36!)^2\bmod 73$.

$$36! = 2^{34}×3^{17}×5^8×7^5×11^3×13^2×17^2×19×23×29×31 $$ $$ \equiv 55\times 24\times 2 \times 17 \times 17 \times 23\times 70 \times 19\times 23\times 29\times 31 $$ $$ \equiv 27 $$Then since $27^2 = 729\equiv -1$, we have $\binom{72}{36} \equiv -1/-1 =1$.

Integrand
  • 8,457
  • You actually calculated that? How long did it take approximately? I tried doing that (before having any knowledge of modular arithmetic, infact these type of questions motivated me to study modular arithmetic myself which is not taught) – UmbQbify Jun 20 '20 at 21:02
  • 2
    About 5-10 minutes; it's a Saturday and I don't have too much better to do. Repeated squaring helps a lot, as well as cute tricks (for instance, $55\cdot 4 = 220\equiv 1$, $2^9 =512\equiv 1$, $3^6=726\equiv -1$) that speed things up. – Integrand Jun 20 '20 at 21:04
  • 2
    There are many very intelligent users on this website, so I won't be surprised if someone finds a more clever solution. – Integrand Jun 20 '20 at 21:05